Contrarian on the Census

I can't resist being a contrarian on the census.  We liberals dislike the decision to include a question on citizenship in the 2020 census.  The fear is that such a question will increase fear of the government among immigrants, particularly those who are undocumented ("illegal").   That fear seems reasonable. The result would be that immigrants would be undercounted.  Because population counts are the basis for determining electoral districts, immigrants would be underrepresented, and because the counts also are used for distribution of government dollars under some programs, immigrants would suffer.

I've no problem with this logic, so what's my contrarian take?

The Trump administration points out that the American Community Survey (an effort conservatives once tried to scuttle IIRC) has always asked about citizenship. The survey gets very detailed, and its results are used in sociological research and government programs.

My contrarian question is this: if immigrants would be fearful of government questions about citizenship, aren't they already fearful of the ACS?  If so, doesn't that impair the validity of the survey?  And if so, why hasn't the Census Bureau fixed the problem?  And if they have, why wouldn't that work for the 2020 census? 

If the problem can't be fixed, do the users of the ACS know of the distortion?